Skip to main content

Task 02: Hardware Discovery via Dell Redfish API

Runbook Azure Dell

DOCUMENT CATEGORY: Runbook
SCOPE: Azure Local hardware provisioning
PURPOSE: Collect complete hardware inventory, BIOS configuration, and iDRAC configuration from all cluster nodes via Redfish API
MASTER REFERENCE: Dell Redfish API Guide

Status: Active

Overview​

Discover all cluster node hardware by connecting to each node's iDRAC via the Redfish API from the management server. This task collects everything needed for Tasks 03–05 in a single execution: hardware inventory, complete BIOS configuration, and complete iDRAC configuration.

Data collected per node:

  • Service tag, model, BIOS version, firmware versions
  • iDRAC MAC address
  • System NIC MAC addresses
  • All BIOS attributes (~256 settings) — used in Tasks 04 and 05
  • All iDRAC attributes (~1024 settings) — used in Tasks 04 and 05
  • CPU, memory, storage inventory
No Node Access Required

All scripts in this task run from the management server and connect to iDRAC over the network via HTTPS (port 443). No access to the nodes themselves is required.


Prerequisites​

RequirementDetailsSource
iDRAC IPsAll nodes reachable on port 443nodes[].bmc.ip_address in infrastructure.yml
iDRAC CredentialsAdmin username and passwordKey Vault or infrastructure.yml
Network AccessManagement server can reach iDRAC OOB networkPhase 03 Network Infrastructure
PowerShell 7+Invoke-RestMethod and powershell-yaml moduleManagement server

Values from infrastructure.yml​

The script reads these keys from infrastructure.yml (schema 5). Each key, with its type, default and example, is in the variables reference.

nodes[]: name, bmc.ip_address, bmc.type, bmc.username, bmc.password, serial_number
hardware.bmc: username, password, verify_certificate
hardware.inventory: output_directory, file_name_prefix, format, timeout_seconds
hardware.vendor, hardware.model, hardware.cpu_cores, hardware.memory_gib, hardware.gpu (enabled, model, count_per_node),
hardware.drives (capacity and cache: media, count, size_tb): compared with what the BMC reports
secrets, key_vaults: read by the secret resolver for the references above

Execution Options​

When to use: Single nodes, ad-hoc verification, or when scripted access is unavailable

Procedure​

  1. Navigate to https://<idrac-ip> and log in with iDRAC credentials
  2. System Information: Dashboard → record Service Tag, Model, BIOS Version
  3. iDRAC MAC: iDRAC Settings → Network → record MAC Address
  4. System NICs: System → Network Devices → record MAC addresses for all ports
  5. Export config: Configuration → Export Server Configuration Profile → JSON format
  6. Repeat for each node

Validation​

  • Service tag recorded for all nodes
  • iDRAC MAC addresses recorded for all nodes
  • System NIC MAC addresses recorded for all nodes
  • Configuration profiles exported and stored

Update infrastructure.yml​

After collecting data from all nodes, manually update the following fields in infrastructure.yml:

FieldSourceWhere to find it
nodes[].serial_numberService TagDashboard → System Information
nodes[].network_adapters[].mac_addressThe MAC address of each port (the management port first)System → Network Devices → MAC for each port
nodes[].network_adapters[].nameNIC adapter nameSystem → Network Devices → adapter name (e.g., Slot 3 Port 1 or Embedded NIC 1)
Save and commit infrastructure.yml after updating

All subsequent tasks depend on these values. Do not proceed to Task 03 until infrastructure.yml is updated and committed.


Run It in a Pipeline​

The toolkit has a discovery pipeline on each platform: pipelines/github-actions/discovery.yml, pipelines/gitlab/stages/discovery.yml (include it from a pipeline file of its own, not from .gitlab-ci.yml) and pipelines/azure-devops/discovery.yml. Each runs scripts/common/discovery/Invoke-AzlDiscoveryRun.ps1 on the general self-hosted runner and nothing else. The same command runs by hand:

# The BMCs (any maker), the Azure tenant and the Active Directory OU, into one folder
./scripts/common/discovery/Invoke-AzlDiscoveryRun.ps1 -InfrastructureFile ./infrastructure.yml -OutputDirectory ./discovery-run

# The Dell iDRAC discovery, and a copy of infrastructure.yml with the serial numbers and MACs found
./scripts/common/discovery/Invoke-AzlDiscoveryRun.ps1 -InfrastructureFile ./infrastructure.yml -OutputDirectory ./discovery-run -Scope DellDiscovery -WriteBack
Pipeline scopeScript parameter -ScopeWhat runs
all (default)HardwareInventory, AzureTenant, ActiveDirectoryThe three below
hardwareHardwareInventoryGet-HardwareInventory.ps1: every BMC over the standard Redfish API, any maker
dellDellDiscoveryInvoke-HardwareDiscovery.ps1: Dell iDRAC only, with the proposal for nodes[]
tenantAzureTenantInventory-AzureTenant.ps1: the Azure tenant of azure.tenant_id
directoryActiveDirectoryGet-OnPremInventory.ps1: the computer objects of the Active Directory OU of the instance
  • Read only, so no approval. The BMCs get GET requests only; Azure and Active Directory are only read.
  • The report. discovery-report.json (what ran, the exit code of each step, the files each step wrote, the verdict), summary.md and a folder per step. The pipeline publishes them as the artifact discovery-report, also when a step fails. The run fails when a step fails. Exit codes of the script: 0 every step succeeded, 1 at least one step failed (the others still ran), 2 bad input (nothing was run).
  • The file must be readable YAML. Before it creates or runs anything, the script reads infrastructure.yml as a YAML mapping. A file that cannot be read that way ends the run with exit code 2: no step runs and no folder is created. The file is not checked against the schema here, because a file that is still being filled in is what discovery is for.
  • The write-back edits a copy only. With the pipeline input write_back (GitLab DISCOVERY_WRITE_BACK, scope dell), Update-InfrastructureYml-FromDiscovery.ps1 -Apply runs on a copy of the master file in the artifact (write-back/, with write-back/changes.txt). The file in the repository is never changed, and no pipeline commits, pushes or opens a pull request. Review the copy and commit it through a pull request of your own.
  • The BMC login is the secret reference in infrastructure.yml (hardware.bmc, or nodes[].bmc). The toolkit secret resolver reads it from Key Vault at run time with the Azure sign-in of the job. No pipeline file or variable holds a password.
  • The runner needs a route to the BMC addresses (HTTPS, TCP 443) and, for the Active Directory step, the ActiveDirectory module (RSAT) on a domain-joined runner.

Platform details: GitHub Actions, GitLab CI/CD, Azure DevOps.


Output Files​

FileWritten byLocation
<prefix>-<UTC time>.json (or .yml): the inventory of every nodeGet-HardwareInventory.ps1hardware.inventory.output_directory (default hardware-inventory)
<service-tag>.json per node, discovery-summary.json, infrastructure.discovered.yml (the proposal for nodes[])Invoke-HardwareDiscovery.ps1-OutputPath (default .\discovery)
discovery-report.json, summary.md, a folder per stepInvoke-AzlDiscoveryRun.ps1 (the pipeline)-OutputDirectory

Keep these folders out of git: they hold serial numbers and MAC addresses.


Troubleshooting​

IssueCauseResolution
Connection refusediDRAC not accessibleVerify iDRAC IP and port 443 reachable from management server
Authentication failedIncorrect credentialsVerify iDRAC password; check root account is enabled
Certificate errorSelf-signed certEnsure PowerShell 7+ and -SkipCertificateCheck is used
TimeoutSlow iDRAC responseCheck iDRAC is responsive via web UI; increase script timeout


Alternatives​

The procedures in this task use the methods shown in the tabs above. The toolkit is PowerShell 7 only and has no Bash scripts. A few tasks that act on Azure also have an Azure CLI script written in PowerShell. The script status page of the toolkit lists which task has which script.

← Task 01: iDRAC DHCP Reservations↑ Phase 01: Hardware ProvisioningTask 03: Management NIC Reservations →

Version Control​

VersionDateAuthorChanges
1.02026-01-01Azure Local CloudInitial document
1.12026-03-03Azure Local CloudFix tab labels, schema keys, standards alignment
1.22026-03-04Azure Local CloudAdd infrastructure.yml update instructions to all three tabs; document Update-InfrastructureYml-FromDiscovery.ps1 integration
1.32026-11-16Azure Local CloudThe discovery run refuses a file that cannot be read as YAML (exit code 2, nothing runs)
1.32026-11-15Azure Local CloudAdd the discovery pipeline (GitHub Actions, GitLab CI, Azure DevOps) and the write-back to a copy; correct the output files